The server defines 31 read-only tools with consistent naming (all start with 'get_'), detailed descriptions, and proper input schemas. Most tools follow the verb_noun pattern and include parameter descriptions with types. However, there are notable gaps: no output schemas are documented (only inferred from response model classes), no error handling guidance is provided, and most tool descriptions lack actionable context about when to use each tool vs. alternatives. Tool compositions are granular (31 separate reads) but descriptions don't explain dependencies or multi-step workflows. Parameter constraints are present in descriptions but not always formalized as enums. The server uses Pydantic response models (SolanaBalanceResponse, etc.) which suggests structured output, but these schemas are not visible in the source excerpt and cannot be scored directly.

• Document output schemas for all 31 tools. Create a schema reference (e.g., in README or OpenAPI spec) showing field names, types, and what each field contains. Example: 'get_solana_balance returns {balance: number (lamports), blockhash: string, timestamp: ISO8601}'.
• Add error handling guidance to all tool descriptions. Example: 'If address is invalid, returns error with format requirement. Validate 44-character base-58 string before calling.' This enables agent self-correction.
• Add readOnlyHint=true to all 31 tool definitions. This signals to agents that these are stateless, safe-to-retry queries.
• Replace free-form string parameters with enums. Convert commitment parameter to enum ["processed", "confirmed", "finalized"]. Convert encoding to enum ["base58", "base64", "base64+zstd", "jsonParsed"]. Convert transaction_details to enum ["full", "accounts", "signatures", "none"].
• Document pagination. For tools returning lists (get_blocks, get_program_accounts, get_leader_schedule, get_largest_accounts, get_inflation_reward), state the result cap and whether pagination is supported. Example: 'Returns up to 100 results. For larger datasets, use start_slot/end_slot range parameters.'
• Enhance descriptions for zero-parameter tools (get_health, get_identity, get_genesis_hash, etc.). Add context like 'Call this first to verify node connectivity before querying account data' or 'Returns the cluster's current configuration; useful for validating RPC endpoint behavior.'
• Add chaining hints. Document common multi-step workflows in tool descriptions. Example in get_account_info: 'After retrieving account info, use get_inflation_reward to check if account is a validator.'